When I use this simple query in the ADF Copy Data tool it returns ~120 columns out of the 210 that exist in the CRM entity. When I sort the data by [createdon] in descending order it returns 137 columns, including several of the new ones that were recently added by our CRM admin.

This leads me to believe that ADF is sampling data to generate the schema and if in the Top X that is pulled, an entirely empty column will not be included for mapping. Even if I write a query that specifies each of the 210 columns individually, the null ones are not returned for mapping.

Can I force ADF to pull the entire dataset for the Import Schemas tool somehow?

<fetch>
<entity name='opportunity'>
<all-attributes/>
<order attribute='createdon'/>
</entity>
</fetch>